Defenses of induction Notes for October 13

Main points

We discussed one broad way of defending inductive inferences against Hume’s argument. Hume’s conclusion, of course, is bizarre: our observations about the past give us no reasons for drawing any particular conclusions about the future. We ended by considering whether pragmatic reasons, namely, that we couldn’t get by without inductive inferences, could be enough to answer Hume. We will take up this question again when we consider Hume’s so-called “skeptical solution” to his problem.

Hume’s steps

I said Hume takes two steps.

1. Claim that we need a principle to move from observations to conclusions about the future (see p. 22).** Strictly speaking, the conclusions are about things we have not observed, whether they are in the past, present or future. This principle is the Uniformity Principle: “that the future will resemble the past” (p. 24) or, “that the course of nature will continue uniformly the same” (p. 130).
2. Deny that we have any reason to believe the Uniformity Principle. (see p. 22–4 for his argument).

The defense of induction we discussed tried to block Hume’s argument at the first step. It denied that we need a principle to move from observations to conclusions.

Do we need a principle?

The challenge to Hume begins with an observation that the reason he gave to explain why we need the principle does not appear compelling. Hume seems to hold inductive reasoning to the standards of deductive reasoning when he says that without the principle we could not be certain of the conclusion.

The challenge then moved from this point to the assertion that we can simply move from observations to conclusions without any principle at all to connect them.

I defended Hume on the grounds that we do need a principle to identify which observations are relevant to the conclusions we wish to draw. If we did not have one, we would draw the wrong inferences in the cases of the Men’s Club and my failure to visit Arkansas. I said that our need for this principle does not rely on assimilating inductive and deductive reasoning. It’s a part of our normal standards for good inductive reasoning: we need to have some standard of relevance to tell us which observations might give us reason to draw which conclusions.

Once we admit that we need a principle of relevance, though, Hume’s problem comes roaring back. For surely part of that principle is the assumption that nature is uniform, such that it is the same for the unobserved future as well as the observed past. If that were not so, our observations of the past would not support our conclusions about the future.